Research And Application Of Community Structure Discovery In Multi-relational Social Networks

With the development of information technology,large-scale social networking sites such as Facebook and Twitter have arisen.The virtual network world is rapidly integrated into people's daily lives.The analysis and mining of these social networks can display important information such as user behavior patterns and interest preferences.Because of the great commercial value and research significance,it has attracted the attention of more and more researchers.Community structure is one of the most important properties in social networks.The feature is the connections between nodes in same community are relatively close and connections between nodes in different communities are relatively sparse.Discover the community structure in the network is an important way to understand the hierarchical structure and function of the entire network.On the other hand,the study of social networks has also promoted the development of other areas such as recommendation systems,and is often used to alleviate sparse problem and cold start issues in the recommendation process.However,current community structure discovery algorithms are mostly based on a single relationship in the network,but there are multiple relationships among individuals in the social network.These relationships determine the community structure in the network together,in order to accurately discover the community structure in the multi-relational social network,and improve the performance of the recommendation system combined with the research of social networks,this article conduct in-depth study of relevant knowledge,the main studying contents and major research achievements are as follows:1.Community structure discovery in the network is often regarded as clustering on complex networks.In order to divide similar nodes in the network into the same community,this paper proposes a community structure discovery algorithm by studying the propagation of information on multi-relational social networks.First,transforms the nodes in the network into vector form through the process of information dissemination.Each group of vectors represents the influence of the node on other nodes in the network,and then the clustering algorithm is used to process these vectors to obtain different Community structure,the nodes in the same community have similar influence in the network,and relations between nodes in same community are relatively closer.2.The traditional collaborative filtering algorithm is directly based on the entire user commodity scoring matrix.The increase of user and item data tends to reduce the efficiency of the algorithm.In order to solve this problem,this paper proposes acollaborative filtering algorithm based on the use of multi-relational social networks.The community structure detection algorithm is used to find communities in social networks,and then predict score of target user based on social network characteristics between community members,finally recommend Top-N products to object user to reduce the time complexity.Experiments based on the Epinions data set at the end of the paper show that this improved algorithm can greatly improve the recommendation efficiency while retaining the accuracy of results.